LLM_MODEL and LLM_EMBEDDING_MODEL environment vars, closes #932

docs/embeddings/cli.md

@@ -31,6 +31,13 @@ The `llm embed` command returns a JSON array of floating point numbers directly
 ```
 You can omit the `-m/--model` option if you set a {ref}`default embedding model <embeddings-cli-embed-models-default>`.
 
+You can also set the `LLM_EMBEDDING_MODEL` environment variable to set a default model for all `llm embed` commands in the current shell session:
+
+```bash
+export LLM_EMBEDDING_MODEL=3-small
+llm embed -c 'This is some content'
+```
+
 LLM also offers a binary storage format for embeddings, described in {ref}`embeddings storage format <embeddings-storage>`.
 
 You can output embeddings using that format as raw bytes using `--format blob`, or in hexadecimal using `--format hex`, or in Base64 using `--format base64`:

docs/usage.md

@@ -30,8 +30,13 @@ Or if you know the name is too long to type, use `-q` once or more to provide se
 ```bash
 llm 'Ten names for cheesecakes' -q 4o -q mini
 ```
+To change the default model for the current session, set the `LLM_MODEL` environment variable:
+```bash
+export LLM_MODEL=gpt-4.1-mini
+llm 'Ten names for cheesecakes' # Uses gpt-4.1-mini
+```
 
-You can also send a prompt to standard input, for example:
+You can send a prompt directly to standard input like this:
 ```bash
 echo 'Ten names for cheesecakes' | llm
 ```

llm/cli.py

@@ -292,7 +292,7 @@ def cli():
 @cli.command(name="prompt")
 @click.argument("prompt", required=False)
 @click.option("-s", "--system", help="System prompt to use")
-@click.option("model_id", "-m", "--model", help="Model to use")
+@click.option("model_id", "-m", "--model", help="Model to use", envvar="LLM_MODEL")
 @click.option(
     "-d",
     "--database",
@@ -783,7 +783,7 @@ def prompt(
 
 @cli.command()
 @click.option("-s", "--system", help="System prompt to use")
-@click.option("model_id", "-m", "--model", help="Model to use")
+@click.option("model_id", "-m", "--model", help="Model to use", envvar="LLM_MODEL")
 @click.option(
     "_continue",
     "-c",
@@ -2352,7 +2352,9 @@ def uninstall(packages, yes):
     type=click.Path(exists=True, readable=True, allow_dash=True),
     help="File to embed",
 )
-@click.option("-m", "--model", help="Embedding model to use")
+@click.option(
+    "-m", "--model", help="Embedding model to use", envvar="LLM_EMBEDDING_MODEL"
+)
 @click.option("--store", is_flag=True, help="Store the text itself in the database")
 @click.option(
     "-d",
@@ -2494,7 +2496,9 @@ def embed(
     "--batch-size", type=int, help="Batch size to use when running embeddings"
 )
 @click.option("--prefix", help="Prefix to add to the IDs", default="")
-@click.option("-m", "--model", help="Embedding model to use")
+@click.option(
+    "-m", "--model", help="Embedding model to use", envvar="LLM_EMBEDDING_MODEL"
+)
 @click.option(
     "--prepend",
     help="Prepend this string to all content before embedding",

tests/test_llm.py

@@ -624,6 +624,18 @@ def test_schema(mock_model, use_pydantic):
     assert response.prompt.schema == dog_schema
 
 
+def test_model_environment_variable(monkeypatch):
+    monkeypatch.setenv("LLM_MODEL", "echo")
+    runner = CliRunner()
+    result = runner.invoke(
+        cli,
+        ["--no-stream", "hello", "-s", "sys"],
+        catch_exceptions=False,
+    )
+    assert result.exit_code == 0
+    assert result.output == "system:\nsys\n\nprompt:\nhello\n"
+
+
 @pytest.mark.parametrize("use_filename", (True, False))
 def test_schema_via_cli(mock_model, tmpdir, monkeypatch, use_filename):
     user_path = tmpdir / "user"